Welcome, Commercial Drone Pilots!
Join our growing community today!
Sign up

Using the DJI Spark for mapping

jefferson3

Member
Joined
Jan 5, 2018
Messages
23
Reaction score
15
Age
44
Location
VA
Hi all!

TL : DR - I am looking into using a "selfie" drone to do GIS mapping (not surveying). Follow along with my adventures if you so choose! ***EDIT*** Just want to add, this thread is not a discussion on if there are better options than a Spark (of course there are) so you can keep those comments to yourself. I have seen people use a ballon for mapping, so a Spark is way ahead of that.

Long version:

As mentioned in my introduction to the site I am new to drones but not to the data created by them. Over the years I have spent many hours processing and utilizing aerial and satellite photographs, many of those hours dedicated to extracting information for analysis.

That said, I know a lot of the commercial mapping drones have been the larger and higher end models (like the DJI Inspire or even the Phantom 4 Pro), but what about the new crop of small drones that are still packed with sensors?

I got a DJI Spark for Christmas and because I am a GIS nerd the first thing that went through my mind was, "can I collect data to view in GIS"? After some research it seemed as though it should be technically possible but there are definitely some pros and cons and I just wanted to document my journey here.

Pros:
  • Super small
  • Portable
  • Inexpensive (comparatively)
  • Easy* to fly (this is subjective of course, but the bar is lower than many professional quads)
Cons:
  • Doesn't support way point flights (through DJI apps at least)
  • Lack of redundant sensors
  • Shorter flight times
  • Shorter transmission distance (than the newer drones like the Mavic)
After putting the Pro/Con list together it looks like the Pros make it a potential candidate for people that might not have deep pockets or want to capture small amounts of data in remote places where size is essential. Next I decided to see if the Cons could be addressed.

1. Doesn't Support way point flights.

While this is technically true, I did find that an iOS app called Autopilot (from Hangar) does support way point flights on the Spark. For this to work though it requires that the aircraft be connected to the RC for the entirety of the flight or else it will RTH without finishing the mission.

2. Lack of redundant sensors

While there really isn't any way to address this by adding more hardware (I don't think this thing could take much of a payload at all seeing as how it only weighs around 300 grams) there is the ability to be conscious of this and pay more attention to your logs and errors to make sure you don't push your luck and one of the sensors goes out while you are in a bad place.

3. Shorter flight times

This one is a bit of a pain, but the easy answer is to just buy more batteries. If you need to fly a longer mission it looks like the Autopilot software will enable to you start and stop at "checkpoints" throughout the campaign. Breaking up a larger mission will allow you to change batteries, allow things to cool down (or heat up depending on climate), and readjust your ground position if needed.

4. Shorter transmission distance

Short transmission distance in this case means 1.2 miles (in ideal conditions), so in reality this will probably never really be an issue since the short battery life will dictate that you can't fly that far away to begin with. But if it does every become an issue then there are ways to boost your signal. I have heard that people get fairly good results with a simple parabolic reflector on the RC. That is on my list of things to test if I ever run into signal issues.


So, where do I go from here? Right now it is freezing outside and I am still in the research phase. I plan on creating missions in Autopilot to collect photographs with a high level of overlap and then feed the images into Drone2Map from Esri. From there I will see what kind of outputs I can get, from just normal orthophotographs to point clouds and 3D models. Just will have to wait and see.

Let me know if you have any (relevant) thoughts, questions, or suggestions.

Jeff
 
Last edited:
I will be interested to see how this goes for you. Good luck with your venture and please do post back with the results. I would definitely worry that the limitations of the Spark will hinder the results. I would think that the reduced transmission range will get in the way in most real-world scenarios. 1.2 miles as you mentioned is in super-ideal conditions which are likely not what will be experienced in real missions.And the shorter battery life will likely drive you crazy when trying to capture large amounts of data. (And I say that lovingly as I love my Spark for what it does. I got it for exactly the "pros" that you mention.)

Again, good luck. I will be following along.
 
I was disappointed with the lack of automated support for the Spark as well, but understand its limitations. The Mavic has been a decent compromise for me, still small to take it just about wherever I go and still be able to do a lot of the stuff my bigger drones do.
 
I will be interested to see how this goes for you. Good luck with your venture and please do post back with the results. I would definitely worry that the limitations of the Spark will hinder the results. I would think that the reduced transmission range will get in the way in most real-world scenarios. 1.2 miles as you mentioned is in super-ideal conditions which are likely not what will be experienced in real missions.And the shorter battery life will likely drive you crazy when trying to capture large amounts of data. (And I say that lovingly as I love my Spark for what it does. I got it for exactly the "pros" that you mention.)

Again, good luck. I will be following along.

Thanks! I am trying to keep my expectations low overall and will try to do my best to mitigate short battery life with better planning and smaller missions, but I do plan on testing the limits as well to see what it can accomplish.
 
I was disappointed with the lack of automated support for the Spark as well, but understand its limitations. The Mavic has been a decent compromise for me, still small to take it just about wherever I go and still be able to do a lot of the stuff my bigger drones do.

The Mavic is definitely a better platform for testing something like this, but I decided to take the hard route! Also, since I got the Spark I must say that being able to palm launch and land has been great. I can see that being very useful in remote places where a flat/open landing area is difficult to find.

I am also happy that the Spark's battery can be charged while in the aircraft over USB with a power bank. Again, just a nice feature, but it all adds up to something that "normal" people can deal with easier.

Over on the SparkPilots forum there are a lot of people hoping for some of the automated functions to be enabled on the Spark and some have even gone as far as using "modded" DJI Go 4 apps. I am not sure if I am there yet though. We'll see how Autopilot does first.
 
Interesting thoughts here! I started mapping with drones as a supplement to my land surveying company. I started with a phantom 3 pro and then moved up to the inspire 1 pro with the X5 camera. The phantom delivered good results and the spark/mavic share the same size sensor which means they might be capable as well. However, I happened to pick up a mavic and did some tests with it to see if it was a viable option for smaller jobs and ran into a few issues. First was the jpg compression, at an altitude above around 100' there is significant pixel bleeding(like a water color effect) which affected accuracy. The raw dng images fixed this but from what I understand the spark doesn't support dng format. However this was not the case with the jpegs shot with the phantom or inspire which means that dji uses a different compression profile for each aircraft so perhaps the spark doesn't have this problem. The second issue I hit was that the gimbal is very inconsistent and seems to only point at a true nadir(straight down) of You fly right after a gimbal calibration on a level surface, after a couple flights it would be pointing off in inconsistent directions. This also affected the accuracy of the resulting orthomosaics as much as a foot horizontally. Now inknow the gimbal is completely different on the spark so again, may not be an issue. I think in the end whether or not it is a viable option might depend on what type of accuracy your looking to achieve.

As for the flight planning you could try the dji GS pro app, the spark is not on the list of supported aircraft but neither is the mavic and it works with mine.
 
  • Like
Reactions: dempseycreekkid
Interesting thoughts here! I started mapping with drones as a supplement to my land surveying company. I started with a phantom 3 pro and then moved up to the inspire 1 pro with the X5 camera. The phantom delivered good results and the spark/mavic share the same size sensor which means they might be capable as well. However, I happened to pick up a mavic and did some tests with it to see if it was a viable option for smaller jobs and ran into a few issues. First was the jpg compression, at an altitude above around 100' there is significant pixel bleeding(like a water color effect) which affected accuracy. The raw dng images fixed this but from what I understand the spark doesn't support dng format. However this was not the case with the jpegs shot with the phantom or inspire which means that dji uses a different compression profile for each aircraft so perhaps the spark doesn't have this problem. The second issue I hit was that the gimbal is very inconsistent and seems to only point at a true nadir(straight down) of You fly right after a gimbal calibration on a level surface, after a couple flights it would be pointing off in inconsistent directions. This also affected the accuracy of the resulting orthomosaics as much as a foot horizontally. Now inknow the gimbal is completely different on the spark so again, may not be an issue. I think in the end whether or not it is a viable option might depend on what type of accuracy your looking to achieve.

As for the flight planning you could try the dji GS pro app, the spark is not on the list of supported aircraft but neither is the mavic and it works with mine.

Thanks for the input from your experience! The lack of RAW support does concern me a little and could be on my Con list as well as the potential issues with the gimbal. I guess I am thinking of the Spark in terms of a "quick and dirty" end product with hopefully minimal effort, unless of course it does much better than I expect. My end products would not require a super high level of accuracy so some of the pressure is off, but I am also working directly with Esri to optimize the workflow with the Drone2Map software as well as inputting high accuracy Ground Control Points with our Trimble Zephyr 2. I will be curious to see how the software does in parsing less than optimal images. Out of curiosity, what software were you using to create your orthos?

Once I start flying and getting data I will post more of my results, but be kind as I am definitely not a surveyor and am use to deriving data from some quite suspect imagery and calling it a day. ;-)

I will take a look at the DJI GS Pro app but I am pretty sure that it is a non-starter for the Spark.
 
Thanks for the input from your experience! The lack of RAW support does concern me a little and could be on my Con list as well as the potential issues with the gimbal. I guess I am thinking of the Spark in terms of a "quick and dirty" end product with hopefully minimal effort, unless of course it does much better than I expect. My end products would not require a super high level of accuracy so some of the pressure is off, but I am also working directly with Esri to optimize the workflow with the Drone2Map software as well as inputting high accuracy Ground Control Points with our Trimble Zephyr 2. I will be curious to see how the software does in parsing less than optimal images. Out of curiosity, what software were you using to create your orthos?

Once I start flying and getting data I will post more of my results, but be kind as I am definitely not a surveyor and am use to deriving data from some quite suspect imagery and calling it a day. ;-)

I will take a look at the DJI GS Pro app but I am pretty sure that it is a non-starter for the Spark.

No problem, I'm happy to help any way I can. I use agisoft photoscan pro for processing and use a Topcon GR-3 system for surveying in GCPs. I think if you can get the flight planning figured out for sufficient overlap you should have no problem producing a "quick and dirty" ortho that could serve well for many purposes in which high accuracy isn't required. I still use the mavic on jobs like that. Sometimes my clients just want an up to date view of a site comparable to google maps and for jobs like that it delivers and I can deploy it and pack it up much quicker than my bigger drones.

Feel free to reach out if there's any other way I can help. I primarily use UGCS for flight planning and while the spark isn't supported in that either, I could create a camera profile to match the sparks to help figure out distances between flight grid lines to nail down overlap settings at different altitudes.
 
  • Like
Reactions: dempseycreekkid
I am looking into using a "selfie" drone to do GIS mapping. Follow along with my adventures if you so choose!
Cons:
  • Doesn't support way point flights (through DJI apps at least)
  • Lack of redundant sensors
  • Shorter flight times
  • Shorter transmission distance (than the newer drones like the Mavic)
You should use the selfie drone for what it's good for.
The Spark makes a great toy and selfie shooter but it's just not equipped for mapping.
The range alone would limit you to small sites, if you could run mapping software.
You mention waypoint software but you don't use waypoint software to generate maps.
You need one of the mapping programs like DroneDeploy.
The lack of RAW support does concern me a little and could be on my Con list
Raw images aren't necessary to create good orthophotos.
You can't write such large files at the speed you would have to fly and since the overlapped images are put through a blender to create an orthophoto raw images would be wasted anyway.
 
Last edited:
You should use the selfie drone for what it's good for.
The Spark makes a great toy and selfie shooter but it's just not equipped for mapping.
The range alone would limit you to small sites, if you could run mapping software.
You mention waypoint software but you don't use waypoint software to generate maps.
You need one of the mapping programs like DroneDeploy.

Raw images aren't necessary to create good orthophotos.
You can't write such large files at the speed you would have to fly and since the overlapped images are put through a blender to create an orthophoto raw images would be wasted anyway.

Don't let the "Surveying" part of the "Surveying & Mapping" section mislead you. I have no intention or illusion of creating a super high resolution "survey grade" product from the Spark. I am a GIS professional and when you do what I do for as long as I have you get used to working with whatever you can get your hands on. So yes, while the Spark is technically a "selfie" drone, it is also technically equipped for mapping. It has a camera and it has GPS. That is all you technically need. Just because you can spend more money and time on something else that can do it better doesn't mean that this effort is not valid. It is a learning experience and after all, isn't that all life is? One big learning experience?

Also, I mention waypoint software in terms of a multifaceted system that would include flight planning, flight execution, and data processing. The actual outputs would be done with Drone2Map by Esri (major GIS software company).

So, "should" I use the Spark as it was intended? Sure. But will I? Nah. At least not for now. ;-)
 
  • Like
Reactions: Flightplan
You should use the selfie drone for what it's good for.
The Spark makes a great toy and selfie shooter but it's just not equipped for mapping.
The range alone would limit you to small sites, if you could run mapping software.
You mention waypoint software but you don't use waypoint software to generate maps.
You need one of the mapping programs like DroneDeploy.

Raw images aren't necessary to create good orthophotos.
You can't write such large files at the speed you would have to fly and since the overlapped images are put through a blender to create an orthophoto raw images would be wasted anyway.
If everyone thought like this than no one would be mapping with drones...after all none of the dji platforms were meant for mapping, they're just flying cameras that people started using for mapping. If no one ever did experiments like this we wouldn't be where we are today.
 
  • Like
Reactions: jefferson3
If everyone thought like this than no one would be mapping with drones...after all none of the dji platforms were meant for mapping, they're just flying cameras that people started using for mapping. If no one ever did experiments like this we wouldn't be where we are today.
And where we are today is that you have a choice of free mapping apps that use the Phantom's SDK to do all the planning, fly the mission, shoot the images with uniform overlaps for you.
The Phantom found it's way into photogrammetry very easily because once the SDK was available, it was perfect for the job.

Using a Spark to attempt a mapping exercise is like riding a child's tricycle to work.
You can do it ... but it's never going to do the job properly.
 
And where we are today is that you have a choice of free mapping apps that use the Phantom's SDK to do all the planning, fly the mission, shoot the images with uniform overlaps for you.
Using a Spark to attempt a mapping exercise is like riding a child's tricycle to work.
You can do it ... but it's never going to do the job properly.
10 years ago professionals in the aerial mapping sector that rely on manned aircraft and cameras that cost more than an I2 would have said the same thing about your phantom and some still do.
 
10 years ago professionals in the aerial mapping sector that rely on manned aircraft and cameras that cost more than an I2 would have said the same thing about your phantom and some still do.
That would be true 10 years ago but it doesn't help someone asking today if he should use a Spark for mapping.
The answer should be no because it just isn't equipped to do the job properly and there are many affordable options that are.
 
That would be true 10 years ago but it doesn't help someone asking today if he should use a Spark for mapping.
The answer should be no because it just isn't equipped to do the job properly and there are many affordable options that are.

To be clear I never asked if I should. I am saying that I am testing if I can.

I am open to hear other affordable options though.

Thanks!
 
That would be true 10 years ago but it doesn't help someone asking today if he should use a Spark for mapping.
The answer should be no because it just isn't equipped to do the job properly and there are many affordable options that are.
I don't see anywhere in the post that the OP asked if he should use a spark. He simply stated that he was going to try it and see if it worked. Trying to shoot him down and say he shouldn't try is counter productive and seemingly not what this forum is about. Also your comments about running images "through a blender" and depending on the SDK to plan, fly and shoot for you seems to show you have a very basic understanding of how this process actually works so maybe instead of interjecting negative comments you could simply leave him to his futile exercise and rest easy knowing your better than this. ;)
 
  • Like
Reactions: jefferson3
@Meta4, I appreciate a good conversation, hence the reason I joined this site. But let's not get off topic too much based on preconceived notions of what we think is appropriate when other tools are available. It is not a stretch to use a drone that has GPS and a gimbaled camera to create maps. Again (and hopefully for the last time) I am not trying to replicate survey-grade products, but rather I am interested if the Spark can create anything meaningful or useful for certain cases.

That said, I am also game to be open about other suggestions (and no, not the suggestions of "don't use a Spark", because that isn't really a suggestion is it?) So if I am overlooking something that costs similar to a spark and has similar pros then feel free to share, but maybe in another thread so I can continue adding to this one as I actually do something!
 
To be clear I never asked if I should. I am saying that I am testing if I can.
I was responding to you saying "Let me know if you have any thoughts, questions, or suggestions".

Go ahead and use whatever you like.
Good luck
 
Last edited:
I was responding to you saying "Let me know if you have any thoughts, questions, or suggestions".

Go ahead and use whatever you like.
Good luck

Lol. Sure, that is a thought, a negative and not very constructive one, but it is a thought.

Don’t get me wrong, I completely understand your point of view of why am I trying to fix what isn’t broken in your opinion. There are other options, I’m not saying you or anyone else should use the Spark, but I’m also not content with the answer that “it isn’t equipped to do the job properly”, because it hasn’t really been tested, so we don’t really know what level of proper it can do!

I do worry that being in a Surveying discussion online will attract more of this type of thinking though, so hopefully this exchange will give others more insight on my goal. (Or maybe, MODS, do we need a GIS/Non-Survey Mapping section?)
 
I do worry that being in a Surveying discussion online will attract more of this type of thinking though, so hopefully this exchange will give others more insight on my goal.
Sorry, I completely misunderstood what you meant by .. Let me know if you have any thoughts, questions, or suggestions.
I won't make that mistake again.
 

Members online

No members online now.

Forum statistics

Threads
4,373
Messages
38,036
Members
6,177
Latest member
NevaWindra